The first useful churn warning comes from reviewed signals

Churn prediction is sold as a model and the habit gets skipped.

The first useful warning usually comes from signals the team already has. Failed payment. Card expiry before renewal. Downgrade. Usage drop. Support ticket rise. Champion exit.

A lagging signal tells you what happened. A leading signal tells you what is about to happen. Five percent monthly churn is 46 percent annual churn. That is the lagging number. The leading version is the conversation you have this week with the account that is about to leave.

The weekly review is twenty minutes. State the observed signal. Check context. Choose one action. Set the next review date. Record the outcome.

A score is useful when it makes the review shorter. It is not a replacement. The checklist decides what to verify. The score sorts who to inspect first.

Most teams never run the checklist because it feels administrative. That is the real gap.